we-solve (beta)

Econ-617 · economics · Economics

Merger effects beyond prices

LLM status Open unverified

Statement

Original subject group: Industrial organization and competition

Audit classification: Published research agenda. Quality among surviving products and total available quality are different estimands. The revised index is defined even when no product survives.

Definitions and precise research target

Fix the population of horizontal mergers in United States grocery-product markets closing during a specified five-year cohort. For merger $m$, define two regimes $a\in\{0,1\}$, where $a=1$ implements the merger and $a=0$ keeps the firms independent. Let $Q_m(a,h)$ be an independently audited quality index and $V_m(a,h)$ the number of distinct product variants available in its premerger geographic market at horizon $h$ years. Normalize quality on a common scale and fix the variant definition before seeing outcomes. The primary targets are $\tau_Q(h)=E[Q_m(1,h)-Q_m(0,h)]$ and $\tau_V(h)=E[V_m(1,h)-V_m(0,h)]$, for $h\in\{1,3,5\}$; expectation averages over the defined merger cohort. Include entry, exit and product redesign within each regime, instead of holding the original product set fixed. Identify these effects using comparable markets under an explicit conditional counterfactual-trends assumption, or report sensitivity bounds when that assumption is relaxed. Quality sampling must not depend only on surviving products, since discontinuation is itself an outcome. Quantify whether quality and variety responses differ with premerger substitution and cost complementarities. This specifies a longitudinal empirical agenda: it does not assume that every merger lowers quality, and it distinguishes changes in product count from changes in consumer-valued variety.

Definitions, assumptions and mathematical qualifications. Let each market have a fixed geographic frame and a finite upper bound $J_{\max}$ on product variants. A variant has a preregistered identity rule independent of merger outcomes. Define $V_m(a,h)=\sum_{j\in\mathcal J_m(a,h)}1$. Let $z_j(a,h)\in[0,1]$ be independently audited quality. One well-defined index including exit is $Q_m(a,h)=J_{\max}^{-1}\sum_{j\in\mathcal J_m(a,h)}z_j(a,h)$, with zero for absence; it measures quality provision, rather than the mean quality among survivors. State whether a different index is desired. For a fixed cohort use specified weights $\nu_m$, $\sum_m\nu_m=1$, in both causal contrasts. A conditional counterfactual-trends assumption requires equality of mean no-merger changes between merger and comparison markets given baseline covariates and adequate overlap.

Variants and assumption changes

  1. Editorial, intensive margin. Restrict to a predeclared set of products that exist under both regimes, and estimate quality effects for that principal stratum; its membership is latent and needs additional assumptions.
  2. Editorial, valued variety. Replace product count by a demand-based equivalent-variation index under specified quasi-linear preferences, including new and discontinued products; this requires identified preferences.

References and source audit

Checked source. 24 August 2024 author draft, Section 6, printed p. 37, sixth research direction. Full primary text retrieved. https://web.stanford.edu/~ayurukog/trendsincompetition.pdf.
  1. Carl Shapiro; Ali Yurukoglu. 2024. Trends in Competition in the United States: What Does the Evidence Show?. 24 August 2024 author draft, Section 6, printed p. 37, sixth research direction. https://web.stanford.edu/~ayurukog/trendsincompetition.pdf. DOI: https://doi.org/10.3386/w32762.

Common conventions: Source mathematical conventions

These conventions apply to each entry unless explicitly replaced.

Models and identification. A primitive model $m$ specifies all probability laws, feasible choices, information, equilibrium and measurement rules used by its target. The admissible class $\mathcal M$ is fixed independently of outcomes. If $P_O(m)$ is its observable probability law and $\tau(m)$ the target, its identified set at observable law $P$ is \[ \mathcal I_\tau(P)=\{\tau(m):m\in\mathcal M,\ P_O(m)=P\}. \] Point identification means that this set is a singleton. An empty set signals incompatibility of data and assumptions. Coordinate bounds need not describe the joint set. Bounds are sharp only if every retained target is attainable or arbitrarily approachable by compatible models. Finite bounds require explicit restrictions on unobserved quantities; unrestricted confounding, hidden wealth, extrapolation or arbitrary equilibrium selection can yield uninformative sets.

Causal interventions. A potential outcome $Y_i(p)$ is the outcome under a complete policy regime $p$, including timing, financing, information and market spillovers. The target population and units are fixed before treatment. With interference, use the complete assignment vector or a justified exposure mapping; individual no-interference notation alone cannot identify an economy-wide intervention. A difference of mean potential outcomes does not require the cross-world joint distribution. Conditioning on post-treatment migration, survival, enrollment or employment changes the estimand and needs separate justification.

Statistical statements. An estimator, confidence set, forecast or test specifies a sampling law, model class and loss/coverage criterion. Uniform $(1-\alpha)$-coverage means $\inf_{m\in\mathcal M}\Pr_m\{\tau(m)\in C_n\}\geq1-\alpha$, or an explicitly asymptotic version. The whole parameter space trivially has coverage, so informativeness requires a stated width, risk or power objective. A forecasting improvement is relative to a named benchmark, information set and evaluation population.

Equilibrium and optimization. An equilibrium comprises feasible plans, optimality, beliefs where needed, budget constraints and all market/resource clearing. The primitives or a stated benchmark define each correspondence $\mathcal E(p;m)$. Nonemptiness is assumed or proved, never inferred just from compact policy sets. A selected-equilibrium target uses a declared selection rule; a robust target quantifies over all permitted equilibria. Use suprema or infima unless attainment is established. An $\varepsilon$-optimal policy is feasible and within $\varepsilon$ of the finite optimum value. Report an empty feasible set separately.

Welfare and accounting. Normative utility, interpersonal normalization, social weights, numeraire, discounting and the use of public receipts are primitives. Behavioral utility need not coincide with the welfare criterion. Household welfare includes ownership income and rents once; adding profits again to compensating variation generally double-counts them. A monetary equivalent is defined by an explicit transfer and price convention, with monotonicity and range conditions. Average effects, mechanism contributions, transfers and welfare losses are different targets. Infinite sums require convergence and dynamic feasible plans require terminal or no-Ponzi conditions.

Source versions. A working-paper date, revision date and journal publication year are distinguished. A DOI for a journal version is not automatically the DOI of an earlier manuscript. Locators refer to the stated version and printed pages where specified. A metadata-only check does not verify a claim about a full-text conclusion. Every entry's source subsection narrows the provenance claim accordingly.

Statement licence: CC-BY-4.0 · Source file

Contributions 1

Econ-617-A1 · Exploration · LLM-generated · gpt 6 astra ultra ·

GPT 6 Astra Ultra attempt on Econ-617

Merger quality incentives and sharp bounds beyond surviving products

Catalogue problem: L41-613. Status: Unresolved. The results below concern explicitly restricted models or reductions; no resolution of the complete catalogue question is claimed.

Question and endpoints

The catalogue asks for quality and variety effects for a fixed merger cohort, geography and horizons, including product entry and exit. Shapiro and Yurukoglu [sy] explicitly discuss non-price merger effects and further research. We derive a quality-investment benchmark at unchanged prices and sharp survivor-selection bounds in a no-entry special case. Neither is an empirical estimate for the grocery-merger cohort.

For each market and regime, let quality of an available product be in $[0,1]$, define variety $V$ as its product count, and define provision \[ Q=J_{\max}^{-1}\sum_{j\text{ available}}z_j. \] Absent products contribute zero. This is distinct from mean quality conditional on survival and is defined even if every product disappears. Cohort weights and the geographic frame are fixed across regimes.

Quality competition under fixed prices

There are two products that remain available in this section. Price is fixed by a pre-existing contract or regulation, giving a common per-unit operating margin $m>0$. Quality $q_j\in[0,\bar q]$ is also the audited quality index, where $0<\bar q\leq1$. Demand is \[ D_1=d_0+\alpha q_1-\beta q_2,\qquad D_2=d_0+\alpha q_2-\beta q_1, \] with $\alpha>\beta\geq0$, $d_0>\beta\bar q$, and $2d_0+2(\alpha-\beta)\bar q\leq1$. These restrictions make demands positive and total demand at most one for every feasible quality pair. An outside good absorbs the remainder.

The local demand system can be rationalized by a strictly concave quadratic net utility in quantities. Let \[ B=\begin{pmatrix}\alpha&-\beta\\-\beta&\alpha\end{pmatrix}, \qquad d=(d_0,d_0)^T. \] At the fixed price, maximize $q^Tx-(x-d)^TB^{-1}(x-d)/2$ over nonnegative quantities with total at most one. The positive definiteness of $B$ and the displayed restrictions give optimizer $x=d+Bq$, exactly the stated demands. Thus the demand response is an explicit primitive with a concave consumer representation, rather than an arbitrary correlation between quality and sales.

Before merger, firm $j$ maximizes $mD_j-\kappa q_j^2/2$, $\kappa>0$. After merger, one owner maximizes the sum of operating margins minus $\kappa_M(q_1^2+q_2^2)/2$, where $\kappa_M>0$ allows a change in real quality-development cost. Assume both optima displayed below lie strictly below $\bar q$.

Theorem (The quality-efficiency threshold): The unique premerger quality equilibrium is symmetric with \[ q_D=m\alpha/\kappa. \] The unique joint-profit-maximizing postmerger qualities are symmetric with \[ q_M=m(\alpha-\beta)/\kappa_M. \] The merger raises, leaves unchanged or lowers quality according as \[ \frac{\kappa_M}{\kappa} \quad\text{is less than, equal to, or greater than}\quad 1-\frac\beta\alpha. \] With two products and $J_{\max}=2$, the provision effect is $q_M-q_D$, while the price and variety effects in this restricted example are zero.
Proof: Each independent firm's own objective has derivative $m\alpha-\kappa q_j$ and strictly negative second derivative $-\kappa$. Its unique best response is therefore $m\alpha/\kappa$, independent of the other's quality; the interior assumption makes clipping unnecessary. Joint demand is $D_1+D_2=2d_0+(\alpha-\beta)(q_1+q_2)$. Joint profit is a strictly concave separable quadratic, with each derivative $m(\alpha-\beta)-\kappa_Mq_j$. This gives its unique optimum. Dividing the two positive optima gives $q_M/q_D=(1-\beta/\alpha)\kappa/\kappa_M$, proving the sign condition. With both products present, normalized provision is their mean quality, so the final statement follows. ∎

The term $\beta$ measures the sales one product takes from the other when quality rises. A merged owner internalizes that loss. A sufficient quality-cost efficiency can offset the reduction in this incentive. No conclusion about general consumer welfare is inferred solely from the sign of product count or from a zero price effect. Product discontinuation is held fixed in this example and is addressed as a measurement problem next.

An exact decomposition including exit

For the next results restrict attention to a preregistered universe of $J_{\max}$ product identities and impose monotone availability: $M_j(1)\leq M_j(0)$ for every identity, where $M_j(a)$ indicates availability at the chosen horizon. This excludes merger-induced entry and is stronger than the primary catalogue assumptions. Draw a market using its fixed cohort weight and then an identity uniformly from the universe. Define $M(a)$ and quality $Z(a)$ for this draw, with quality required only when available. Then \[ p_a={\mathbb{P}}(M(a)=1)={\mathbb{E}}[V(a)]/J_{\max},\qquad \mu_a={\mathbb{E}}[Z(a)\mid M(a)=1],\qquad {\mathbb{E}}[Q(a)]=p_a\mu_a. \] Assume $0<p_1\leq p_0$. The always-available principal stratum is $A=\{M(1)=M(0)=1\}$, of probability $p_1$. The eliminated stratum is $E=\{M(0)=1,M(1)=0\}$, of probability $p_0-p_1$.
Proposition (Provision is not a surviving-product average): Let $\tau_A={\mathbb{E}}[Z(1)-Z(0)\mid A]$. If $p_0>p_1$, then \[ {\mathbb{E}}[Q(1)-Q(0)] =p_1\tau_A-(p_0-p_1){\mathbb{E}}[Z(0)\mid E]. \] If $p_0=p_1$, the exit term is zero and the same formula holds without defining the empty stratum's mean. A rise in observed mean quality among available products can coexist with zero quality change for every always-available product and a fall in total quality provision.
Proof: Under monotone availability, all regime-one survivors belong to $A$, so their quality contribution is $p_1{\mathbb{E}}[Z(1)\mid A]$. Regime-zero provision is $p_1{\mathbb{E}}[Z(0)\mid A]+(p_0-p_1){\mathbb{E}}[Z(0)\mid E]$. Subtraction proves the identity. For the last assertion take two original products with qualities $0.8$ and $0.4$. The merger removes the second and leaves the first unchanged. Mean available quality rises from $0.6$ to $0.8$, but normalized provision falls from $0.6$ to $0.4$, while the sole always-available product has zero quality effect. ∎

Sharp bounds for the always-available quality effect

Suppose a randomized regime design, or other explicitly justified identification assumptions, identifies the full marginal distribution of availability and quality under each regime. It does not observe their joint counterfactual distribution. Let $F_0$ be the quality distribution among regime-zero available products and $F_0^{-1}(u)=\inf\{z:F_0(z)\geq u\}$ its generalized quantile. Put $r=p_1/p_0\in(0,1]$ and define trimmed means \[ L_r=\frac1r\int_0^r F_0^{-1}(u)\,du, \qquad U_r=\frac1r\int_{1-r}^1F_0^{-1}(u)\,du. \] These definitions permit atoms through fractional trimming at a cutoff.
Theorem (Sharp principal-stratum interval): Under the preceding marginal information and monotone availability, the sharp identified interval for $\tau_A$ is \[ \boxed{\quad \mu_1-U_r\ \leq\ \tau_A\ \leq\ \mu_1-L_r.\quad} \] Every point of this interval is compatible with the observed marginal laws. When $r=1$, the interval collapses to $\mu_1-\mu_0$.
Proof: Among regime-zero available products, the always-available group has fraction $r$. Let $a(z)\in[0,1]$ denote its conditional membership probability given baseline quality $z$, so $\int a\,dF_0=r$. Its baseline mean is $r^{-1}\int za(z)dF_0(z)$.

A minimizing membership rule selects the lowest $r$ fraction, randomizing at a quality cutoff if needed. To prove this, let $a_L$ denote that rule and $t$ its cutoff. For $z<t$, $a_L(z)=1$ so $(z-t)(a(z)-a_L(z))\geq0$; for $z>t$, $a_L(z)=0$ and the same inequality holds. At the cutoff the product vanishes. Integrating and using $\int(a-a_L)dF_0=0$ shows $\int za\,dF_0\geq\int za_L\,dF_0=rL_r$. Selecting the highest fraction gives the corresponding upper bound $rU_r$ by reversing the argument. Since ${\mathbb{E}}[Z(1)\mid A]=\mu_1$, subtracting yields the interval.

For sharpness, construct principal strata with masses $p_1,p_0-p_1,1-p_0$. Allocate regime-zero survivor qualities to the first two strata according to either extreme trimming rule. This exactly preserves their observed combined distribution $F_0$. In the always-available stratum, independently draw regime-one quality from its observed survivor distribution. Assign absence in regime one to the eliminated and never-available strata. The resulting joint law obeys monotonicity and reproduces both observed marginal laws while attaining the corresponding endpoint. Mixtures of the two constructions preserve all marginals and monotonicity and attain every intermediate mean. If $r=1$, both trimmed means are $\mu_0$. ∎

The theorem is a complete sharp result for the stated no-entry special case. It does not make principal-stratum membership observable, and it does not recover quality for discontinued products by conditioning on the products that happen to survive.

Counterfactual trends and a joint sensitivity region

Actual mergers are not typically randomized. Suppose a comparison-market strategy supplies counterfactual no-merger mean predictions $q_0^*,v_0^*$ at one horizon. Declare sensitivity allowances $\Gamma_Q,\Gamma_V\geq0$ for violations of the required conditional trends. Let observed postmerger means be $q_1,v_1$. Under only these aggregate restrictions, feasible counterfactual means must lie in \[ \mathcal C=\{(q,v):|q-q_0^*|\leq\Gamma_Q, |v-v_0^*|\leq\Gamma_V,\ 0\leq v\leq J_{\max},\ 0\leq q\leq v/J_{\max}\}. \]
Proposition (A coupled identified region from aggregate restrictions): If no further restrictions on counterfactual product distributions are imposed, the sharp aggregate effect region is \[ \{(q_1-q,v_1-v):(q,v)\in\mathcal C\}. \] An empty $\mathcal C$ means the proposed sensitivity restrictions are incompatible with the accounting bounds. Separate coordinate intervals generally lose the restriction $q\leq v/J_{\max}$.
Proof: Every admissible counterfactual obeys the displayed inequalities, proving necessity. Conversely, take any $(q,v)\in\mathcal C$. If $v=0$, then $q=0$ and let every product be absent. If $v>0$, with probability $v/J_{\max}$ make all $J_{\max}$ products present at common quality $qJ_{\max}/v\in[0,1]$, and otherwise make all absent. Expected count is $v$, and expected normalized quality provision is $q$. Couple this counterfactual independently with the observed regime-one distribution. This construction obeys exactly the stipulated aggregate restrictions and realizes the proposed pair of effects. This proposition is separate from, and does not impose, the preceding no-entry monotonicity restriction. ∎

Remaining gaps

The quality-investment model isolates substitution and quality-cost efficiencies at fixed prices; it does not estimate them or endogenize assortment. The sharp trimming result requires no merger-induced entry, while the primary catalogue target includes it. The sensitivity region is sharp only relative to its deliberately limited aggregate restrictions and can be wide. Geographic spillovers, auditing error, product redesign and the comparability of unmerged markets need additional evidence. The complete multi-horizon grocery-merger effects and demand-based valued variety remain unresolved.

Completion Estimate

63%

This is a post hoc, heuristic estimate for the full catalogue target. The attempt proves a quality-efficiency threshold at fixed prices, sharp survivor-stratum quality bounds under no entry, and a coupled sensitivity region for aggregate quality provision and variety. Full multi-horizon merger effects still need audited grocery evidence, credible comparison-market trends, entry and redesign models, geographic spillovers and identified consumer-valued variety.

References

  • [sy] C. Shapiro and A. Yurukoglu (2024), "Trends in Competition in the United States: What Does the Evidence Show?," author draft dated 24 August 2024, Sections 4B and 6, especially printed p. 37. https://web.stanford.edu/~ayurukog/trendsincompetition.pdf. NBER Working Paper 32762, https://doi.org/10.3386/w32762. The primary draft was inspected; the models and sharp-bound proofs above are separately derived, without a novelty claim for the general trimming principle.

Go to contribution →